Presented at the Cochrane Colloquium, Sao Paulo, Brazil, October 23–27, 2007; and the Canadian Association of Emergency Physicians, Ottawa, Ontario, Canada, June 2–6, 2008.
Accuracy of Ottawa Ankle Rules to Exclude Fractures of the Ankle and Midfoot in Children: A Meta-analysis
Article first published online: 2 FEB 2009
© 2009 by the Society for Academic Emergency Medicine
Academic Emergency Medicine
Volume 16, Issue 4, pages 277–287, April 2009
How to Cite
Dowling, S., Spooner, C. H., Liang, Y., Dryden, D. M., Friesen, C., Klassen, T. P. and Wright, R. B. (2009), Accuracy of Ottawa Ankle Rules to Exclude Fractures of the Ankle and Midfoot in Children: A Meta-analysis. Academic Emergency Medicine, 16: 277–287. doi: 10.1111/j.1553-2712.2008.00333.x
A related commentary appears on page 352.
- Issue published online: 6 APR 2009
- Article first published online: 2 FEB 2009
- Received June 17, 2008; revisions received August 22 and September 24, 2008; accepted September 24, 2008.
- Ottawa Ankle Rules;
- pediatric ankle injury;
- systematic review
Objectives: The objectives were to conduct a systematic review to determine the diagnostic accuracy of the Ottawa Ankle Rules (OAR) to exclude ankle and midfoot fractures in children and the extent to which x-ray use could be reduced without missing significant fractures.
Methods: The authors conducted comprehensive searches of electronic databases and gray literature sources. Independent reviewers applied standard inclusion and exclusion criteria. The criterion standard diagnostic test was an ankle and/or foot x-ray or proxy measure to ensure no missed fractures. Standard 2 × 2 tables were constructed. Sensitivities and specificities were pooled using an approximation of the inverse variance; 95% confidence intervals (95% CIs) were calculated using the exact method. Likelihood ratios (LR ±) and diagnostic odds ratios were combined under DerSimonian and Laird random effects model.
Results: A pooled analysis of 12 studies (N = 3,130) identified 671 fractures (prevalence = 21.4%). Ten studies reported Salter-Harris Type I (SH-I) fractures. The pooled sensitivity was 98.5% (95% CI = 97.3 to 99.2), suggesting that the OAR can be used to rule out a fracture. Four of 10 missed fractures were characterized: 1 SH-I, 1 SH-IV, and 2 “insignificant fractures” (either SH-I or avulsion fractures <3 mm). The pooled estimate for rate of x-ray reduction was 24.8% (95% CI = 23.3% to 26.3%; range = 5% to 44%).
Conclusions: The OAR appear to be a reliable tool to exclude fractures in children greater than 5 years of age presenting with ankle and midfoot injuries. Employing the OAR would significantly decrease x-ray use with a low likelihood of missing a fracture.
Approximately 2% of children in the emergency department (ED) present with an acute ankle and midfoot injury.1 In the ED, radiographs (x-rays) are often used to augment the history and physical examination, despite only identifying fractures in 12% to 21% of patients.2,3 The reported use of ankle and foot x-rays for these patients in the ED varies from 64%4 to 100%.5,6 X-rays may result in unnecessary radiation, decreased patient throughput, increased wait times, and increased health care expenditures.7,8
The Ottawa Ankle Rules (OAR) are a set of clinical decision rules that were derived and validated by Stiell et al.9 for use in adults (18 years or older) with acute ankle and midfoot injuries. The ankle component of the OAR states that radiographic evaluation of the ankle is only necessary if there is pain in the malleollar zone and one of the following: bony tenderness along the distal 6 cm of the posterior edge of either malleolus or inability to weight bear four steps both immediately after the injury and in the ED. The midfoot component states that foot radiographs are recommended only when there is pain in the midfoot region and one of the following: bony tenderness at either the navicular bone or the base of the fifth metatarsal or inability to weight bear four steps both immediately after the injury and in the ED (see Figure 1). The OAR are 100% sensitive for ruling out clinically significant fractures in the adult population.9 EDs that have implemented the OAR have consistently shown a decrease in x-ray use, wait times, and health care expenditures.9,10
Numerous studies have attempted to validate the use of the OAR in children. However, widespread use of the OAR in the pediatric clinical field is inconsistent, likely due to three reasons.11 First, obtaining a reliable verbal history and performing an adequate physical examination can be more difficult in a child than an adult. Second, the OAR can only be applied to children able to walk independently prior to the injury. Third, children have anatomic differences compared to adults, namely the presence of growth plates. These growth plates form the weakest part of the bone and are prone to unique fractures classified by the Salter-Harris (SH) system.
In 2003, Bachmann et al.12 conducted a systematic review on the accuracy of the OAR to exclude fractures of the ankle and midfoot in both adult and pediatric populations. In the subgroup analysis for children, they assumed a prevalence of 15% fracture and calculated a posttest probability of fracture of 1.22% (95% confidence interval [CI] = 0.53% to 3.03%) following a negative OAR result. This seems low enough to be useful; however, the CIs are wide, a reflection of the low numbers of children in the eligible studies. Myers et al.11 also conducted a review to examine the accuracy of the OAR in children; however, their review does not include all of the available studies.
The primary objective for the current review was to conduct a systematic review to determine the diagnostic accuracy of the OAR to exclude fractures of the ankle and midfoot in children and the extent to which ankle/foot x-rays could safely be reduced without missing significant fractures.
This was a systemic review of the literature to assess the accuracy and reliability of the OAR in the pediatric population. Prior to conducting the systematic review, a protocol was developed and followed to ensure transparency and reproducibility of the methodology.
The research librarian, in collaboration with the review team, developed specific search strategies for the following electronic databases: MEDLINE (1992–May 2007), Ovid MEDLINE in-process and other nonindexed citations (up to May 2007), Cochrane Central Register of Controlled Trials (1992–May 2007), EMBASE (1992–May 2007), and CINAHL (1992–May 2007). Trials registers (Current Controlled Trials and ClinicalTrials.gov) were searched for additional unpublished trials. Websites for the Society for Academic Emergency Medicine and the Canadian Association of Emergency Physicians were searched for identification of conference proceedings. The National Library of Medicine Gateway and BioMed Central were searched for identification of meeting abstracts. A combination of subject headings and keywords was adapted for each electronic resource (see Data Supplement S1 for search terms, available as supporting information in the online version of this paper). For some databases, these search terms were combined with a pediatric search filter (available at http://www.ualberta.ca/ARCHE/filters.html# pedsmedline).
In addition, reference lists were hand searched and primary authors of selected studies contacted. Key articles were tracked forward using the cited reference search feature in Web of Science and the citation tracker feature in Scopus. The search had no language restrictions. The search is inclusive from 1992 (the year the OAR were introduced) until May 2007.
Two reviewers independently screened all titles and abstracts generated from the searches. We retrieved the full manuscripts for all titles deemed potentially relevant, and two reviewers independently screened them using a standardized checklist inclusion and exclusion form (see Data Supplement S2, available as supporting information in the online version of this article). Disagreements were resolved through discussion or in consultation with a third reviewer.
Studies were included if the OAR were used to diagnose fractures of the ankle or midfoot in children (≤18 years) who presented to the ED with an acute blunt ankle and/or midfoot injury. Toddlers must have started to walk prior to their injury to assess the weight-bearing component of the OAR. Studies including both children and adults were included if the results for patients ≤18 years were reported separately. The criterion standard diagnostic test was an x-ray of the ankle and/or foot or a proxy measure (e.g., telephone follow-up at 14 days) to ensure that no fractures were missed. Studies had to report absolute numbers comparing the OAR with the criterion standard, such that a 2 × 2 table that specified the true-/false-positive and true-/false-negative rates could be constructed. Study designs that were acceptable included prospective or retrospective cohort studies, cross-sectional studies, or case series with more than 10 patients.
One reviewer extracted data using a standard form (Data Supplement S3, available as supporting information in the online version of this paper); a second reviewer checked for completeness and accuracy and entered the data into an electronic database. Data extracted included study design, study setting, participants (number, age, gender), diagnostic tests (type of assessment—foot, ankle, or combined), when OAR were applied and by whom, and results. For abstracts that were included, attempts were made to contact primary investigators to obtain missing data.
For each individual study, we constructed a standard 2 × 2 table noting the true-positive, false-positive, true-negative, and false-negative results. We tested the variation in diagnostic threshold using both Spearman rho correlation and Lettenberg and Moses method.15,16 Sensitivities, specificities, and x-ray reduction rates were pooled using an approximation of the inverse variance approach; 95% CIs were calculated using the exact method. Likelihood ratios (LR ±) were combined under DerSimonian and Laird random effects model.17 Statistical heterogeneity was quantified using the I2 statistic. This measurement describes the percentage of total variation across studies that is due to heterogeneity rather than chance.18 A value greater than 50% was considered to indicate substantial heterogeneity and grounds for not presenting the combined results.19 A priori subgroup analyses were conducted to explore the heterogeneity based on age (<6 years vs. 6 years or older), study design, type of fracture (S-H Type I [SH-I] vs. other), and type of assessment (ankle, foot, or combination). We conducted a priori sensitivity analyses in order to test how robust the test characteristics were relative to different components of methodologic quality (i.e., prospective vs. retrospective data collection, consecutive enrolment, verification bias, and blinding).
Because the OAR are calibrated toward high sensitivity, we were particularly interested in the pooled sensitivity and negative LR, that is, how many times more likely it was to find a negative result among children with a fracture (1 – sensitivity), than among those without a fracture (specificity). Publication bias was assessed using the effective sample size funnel plot and associated regression test for asymmetry.20 Statistical calculations were made using SAS (SAS Institute, Cary, NC), Stata (StataCorp, College Station, TX), and RevMan (Cochrane Collaboration, Oxford, UK).
The electronic and gray literature searches identified 451 references that were screened for inclusion. The full text of 102 studies was retrieved for further review, and 12 studies met the inclusion criteria (Figure 2).3–6,21–28Table 1 describes the characteristics of the 12 included studies. Most studies were published in peer reviewed journals; three studies were available only as abstracts.24,25,28 Contact with the primary authors of the abstracts did not yield any additional data. Four studies applied the OAR prospectively4,23,24,27 and followed up with patients to identify any missed fractures. In eight studies, the OAR were applied retrospectively (i.e., the index test was either completed at the time of assessment by the attending physician but not used to determine if an x-ray should be taken or the OAR were applied to clinical data at a later date to determine which patients would have qualified for an x-ray).3,5,6,21,22,25,26,28 Five studies included injuries to the ankle and foot;4–6,23,27 7 included ankle injuries only.3,21,22,24–26,28 SH-I fractures were defined as insignificant fractures in 1 study,6 not mentioned in 2,3,4 and included as significant fractures in the other 9. The combined 12 studies provided 3,130 patients for analysis. Overall, 671 fractures were identified for a prevalence of 21.4%. Because the data for the midfoot and ankle portions of the OAR were reported separately in only one study,6 the results of this meta-analysis combine both portions of the rule.
|First Author, Year||Country||Design||Inclusion Criteria||Definition of Significant Fracture||Number Enrolled||Age (Years), Mean ± SD|
|Number of Sites and Setting||OAR Application||Exclusion Criteria||Included Fracture(s)||Definition SH-I Fracture||Number Analyzed||Gender (males), n (%)||% Receiving X-rays|
|Al Omar, 20025||Canada||Prospective observational cohort||Blunt ankle or midfoot injury in past 48 hours||Ankle; midfoot||Any fracture of ankle or midfoot diagnosed by radiologist and needing medical attention||80||12 ± 3.0||100|
|Single pediatric ED||Retrospectively by single investigator blinded to physician clinical assessment and final x-ray diagnosis||<6 years; multiple injuries; open fracture; neurovascular compromise; underlying disease predisposing to fracture or sensory abnormalities; hemophilia; being reassessed; referred with x-rays; intoxicated||NR||80||45 (56.3)|
|Boutis, 200121||Canada, United States||Prospective cohort||Healthy children with isolated acute ankle trauma within past 72 hours||Ankle; low- and high-risk fracture||Fracture with low risk of complications: lateral talar or epiphyseal avulsion, distal fibula, SH-I and -II, metaphyseal buckle. All others considered high-risk fractures||607||Median 12.9, Range 3–16||100|
|Two Pediatric EDs||Retrospectively; blinding unclear||<3 or >16 years; x-rayed before assessment; MSK disease; coagulopathy; developmental delay; history of surgery or injury ≤3 months on same ankle; multisystem trauma||Low-risk fracture||607||302 (49.8)|
|Chande, 199522||United States||Prospective consecutive survey||Acute soft tissue or bony injury to ankle||Ankle||Any fracture, avulsion||71||11.8 ± 3.9||95.8|
|Single pediatric ED||Retrospectively by single investigator blinded to final x-ray diagnosis||Foot injuries; received an x-ray||Growth plate involvement||68||46 (67.6)|
|Clark, 20033||United States||Prospective cohort||Acute nonpenetrating traumatic ankle injury||Ankle||Any fracture||195||12.6 Range 2.8–17.9||100|
|Single pediatric ED||Retrospectively; blinding unclear||Neurological impairment; isolated skin injury; previous x-ray; suspected physical abuse; open fracture; metabolic disease; osteogenesis imperfecta; no phone; intoxicated; pregnant||Point tenderness at growth plate; x-ray confirmation of epiphyseal separation from metaphysis; no metaphyseal fragment at any time||160||107 (54.9)|
|Cuello-Garcia, 200423||Mexico||Prospective cohort||Trauma to ankle or foot||Ankle; midfoot||Fragment >3 mm||112||11.2 ± 3.5||NR|
|Single pediatric ED||Prospectively by ED nurse, resident or pediatrician||Injury > 7 days old; multiple injuries; altered consciousness; MSK disease; reassessment; referred for x-ray||NR||111||65 (58.6)|
|Dohin, 200424||France||Prospective cohort||Trauma to ankle||Ankle||NR||160||11.3 Range 3–15||100|
|Single general ED||Prospectively; blinding unclear||Obvious fracture||NR||154||71 (44.4)|
|Gilligan, 200025||United States||Prospective cohort||Acute ankle injury||Ankle||All fractures||182||Median 13 Range 2–17.5||94.5|
|Two pediatric EDs, one general ED||Retrospectively||NR||Clinical exam plus x-ray||182||NR|
|Karpas, 200226||United States||Prospective cohort||Ankle injury within past 48 hours and met 1 of 4 CPP clinical criteria for x-ray||Ankle||All fractures||190||Median 13 IQR 1–14.5||97.4|
|Single pediatric ED||Retrospectively by investigator to evaluate ED nurse accuracy||Open injury; multiple trauma; developmental delay; injury to same site in past 2 weeks or recurrent injury to same ankle; referral with x-ray; no x-ray taken||NR||185||95 (50.0)|
|Libetta, 19994||United Kingdom||Prospective cohort||Blunt ankle and/or midfoot injury||Ankle; midfoot||NR||761||Median 11 Range 1–15||56.8|
|Single pediatric ED||Prospectively by senior house officers||Had not started to walk yet||NR||761||385 (50.6)|
|McBride,199727||Canada||Prospective cohort||Blunt ankle trauma||Ankle; foot||Fragment >3 mm||37||13.2 ±1.3||100|
|Single general ED||Prospectively by part-time family physicians||Open injury; injury >1 week old; pregnant||NR||37||17 (45.9)|
|Pacheco-Fowler, 199928||United States||Retrospective cohort (rules applied after x-ray)||Traumatic ankle injury||Ankle||Fragment >3 mm||115||NR||100|
|Two pediatric EDs||Retrospectively||NR||Involved growth plate||115||NR|
|Plint, 19996||Canada||Prospective cohort||Acute ankle injury within past 48 hours||Ankle; midfoot||Fragment ≥3 mm diagnosed by radiologist||670||Median 12.6 IQR 9.7–14.7||96|
|Two pediatric EDs||Retrospectively by investigator||Multiple injuries; open fracture; SH-I fractures; neurovascular compromise; underlying disease predisposing to fracture or sensory abnormalities; isolated skin injury; in for reassessment; referred with x-rays; intoxicated||NR||670||359 (53.6)|
Overall, the quality of the included studies was satisfactory (see results of QUADAS in Data Supplement S4, available as supporting information in the online version of this paper). In 10 studies, the spectrum of included patients was representative of those on whom the OAR would be utilized; in 2 studies, it was unclear.24,28 Two studies did not describe their exclusion criteria.25,28 In all cases, the reference standard was an ankle and/or midfoot x-ray taken at the initial ED visit at the discretion of the attending physician; those who were not x-rayed were followed up by telephone or asked to return for reassessment.3,4,6,21,23–25 Seven studies stated that the OAR were interpreted by physicians blinded to the radiologists’ diagnosis, and the radiologists were blinded to the OAR assessment.3,5,6,21–23,26 Five studies did not report blinding. Only Libetta et al.4 applied the OAR prospectively and used the results to determine if an x-ray would be ordered. In the other 3 prospective studies, 1 did not report the proportion of patients x-rayed,23 and 224,27 reported that 100% received an x-ray. The OAR assessment was conducted on the clinical data retrospectively in the remaining eight studies (see Table 1). All studies accounted for the entire population enrolled.
Table 2 summarizes the quantitative results for the 12 studies. The Spearman rho correlation between the sensitivities and the specificities was not significant (Spearman rho = −0.265; p = 0.405). This suggests that the OAR were applied consistently across all studies and it was appropriate to pool the estimates. Furthermore, the included studies were homogenous in terms of population, intervention, and design (i.e., methodologic and clinical heterogeneity was minimal). The pooled sensitivity was 98.5% (95% CI = 97.3% to 99.2%; see Figure 3). The sensitivities were consistently high in 10 of the studies (i.e., ≥97%).4–6,21–27 Clark and Tanner3 reported a sensitivity of 83.3% in those 15 years or younger; for those older than 15 years, the sensitivity was 100%. Pacheco-Fowler et al.28 reported a sensitivity of 93.8% for all fractures; for clinically significant fractures (defined as any fracture involving the growth plate or having a fragment >3 mm), the sensitivity was 100%.
|First Author, Year||Total Number of Diagnosed Fractures, n/N (%) [top] Proportion SH-I Diagnosed [bottom]||Sensitivity* (95% CI)||Specificity* (95% CI)||Potential Reduction in X-rays Using OAR, n (%)||Number of Missed Fractures if Using OAR|
|Al Omar, 20025||17/80 (21.3) 3/17 SH-I||100 (81, 100)||30.2 (19.2, 43.0)||19 (23.8)||None|
|Boutis, 200121||161/607 (26.5) Low risk, 107 (87/107 SH-I); high risk, 54 (6/54 SH-I)||100 (97.7, 100)||15.7 (12.4, 19.4)||70 (11.5)||None|
|Chande, 199522||14/68 (20.6) 3/14 possible SH-I||100 (76.8, 100)||31.5 (19.5, 45.6)||17 (25.0)||None|
|Clark, 20033||30/160 (18.8) SH-I included but no data||83 (65.3, 94.4)||50.0 (41.1, 58.9)||70 (43.8)||5|
|Cuello-Garcia, 200423||35/111(31.5) 18/35 SH-I||100 (90, 100)||7.9 (3.0, 16.4)||6 (5.4)||None|
|Dohin, 200424||34/154(22.1) 21/34 SH-I||100 (89.7, 100)||9.2 (4.7, 15.8)||11 (7.1)||None|
|Gilligan, 200025||46/182 (25.3) 17/46 SH-I||97.8 (88.5, 99.9)||21.3 (14.8, 29.2)||30 (16.5)||1|
|Karpas, 200226||31/185 (16.7) 1/31 SH-I||96.8 (83.3, 99.9)||24.7 (18.1, 32.3)||39 (21.1)||1 SH-I|
|Libetta, 19994||59/761 (7.8) SH-I NR||98.3 (90.9, 100)||46.7 (43.0, 50.5)||55 (7.17)||1 SH-IV|
|McBride, 199727||7/37 (18.9) 2/7 SH-I||100 (59.0, 100)||26.7 (12.3, 45.9)||8 (21.6)||None|
|Pacheco-Fowler, 199928||32/115 (16.5) (19 significant + 13 insignificant) SH-I included but no data||93.8 (79.2, 99.2)||18.1 (10.5, 28.0)||17 (14.8)||Two insignificant fractures|
|Plint, 19996||205/670 (30.6) 119/205 SH-I (96 patients had >1 fracture)||100 (98.0, 100)||33.0 (29.0, 37.0)||184 (24.0)||No significant fractures|
The specificities reported in the individual studies ranged from 7.9% to 50% (Figure 3). Owing to substantial heterogeneity, we decided not to pool these results. Despite conducting sensitivity and subgroup analyses, we were unable to determine the possible sources of heterogeneity.
The pooled negative LR was 0.11 (95% CI = 0.05 to 0.26; I2 = 51%), which suggests that the OAR can be used to rule out a fracture. There was significant heterogeneity when assessing the positive LR (I2 = 93.9%); therefore, we did not feel it was appropriate to pool these results. X-ray reduction rates ranged from 5% to 44%, with the pooled estimate of 24.8% (95% CI = 23.3% to 26.3%). There was no evidence of publication bias.
Based on the mean 21.4% prevalence of fractures among the included studies, and the pooled negative LR of 0.11, the posterior probability of fracture given a negative OAR assessment is approximately 2.9%. Applying the OAR to the included pediatric population would result in a missed fracture rate (1 – negative predictive value [NPV]) of 1.2% (95% CI = 0.6% to 2.3%). This is fairly consistent with the estimate by Bachman et al.,12 who reported a missed fracture rate of 1.22% (95% CI = 0.53% to 3.08%), assuming a 15% prevalence of fracture.12
The results of this current meta-analysis provide Level 2 evidence for the use of the OAR in children, based on the broad validation of the OAR applied by a mix of health professionals all trained in the use of the OAR and in multiple clinical settings with varying prevalence of fractures.29 An OAR sensitivity of 98.5% is particularly important in light of findings by Al Omar et al.5 that demonstrated the sensitivity of physicians’ clinical judgment of the likelihood of an ankle fracture was 64%.5
The OAR are widely accepted as a valid clinical decision rule for ankle injuries in adults by emergency physicians.30,31 The utility of a clinical decision rule that is calibrated toward a high sensitivity and high NPV allows physicians to confidently rule out fractures based on the history and physical examination, without requiring x-rays of the injured ankle. The downside to this is a clinical decision rule that has a low specificity and low positive predictive value.
Ten fractures were missed in 3,130 patients included in this review. Of the 4 that were characterized, 1 was a SH-I fracture,26 1 was a SH-IV fracture,4 and 2 were “insignificant fractures” (either SH-I or avulsion fractures less than 3 mm).28 Unfortunately, the one study that reported 5 missed fractures did not provide a description or classification of the missed injuries;3 similarly, Gilligan et al.25 did not describe their 1 missed fracture.
The question of whether the OAR can detect SH-I fractures is important with respect to use in children. The diagnosis of a SH-I is usually clinical, and x-rays are not required to diagnose SH-I fractures of the distal fibula. However, x-rays are frequently done to exclude a more significant fracture. The OAR should detect SH-I injuries, because they are characterized by maximal tenderness and swelling over the growth plate, which is within 6 cm of the posterior edge of either malleolus. At present there is considerable variation in how pediatric emergency physicians choose to manage SH-1 injuries that range from symptomatic treatment only, to below knee casting and follow-up with an orthopedic surgeon.32 A recent study by Boutis et al.33 suggests that a removable brace for these injuries may lead to better functional outcomes.
As with all systematic reviews, our study has some potential limitations. First, although we conducted a comprehensive search of the literature and did not exclude studies in terms of language or publication status, it is possible that our search strategy did not identify all potentially relevant articles. Second, the range of the specificities (7.9% to 50%) in the included studies is substantial, an observation that was also made in a meta-analysis of the OAR in adults.12 The reasons for this variation are not discernible based on the information provided in the studies, but require further investigation to determine if this inconsistency is based on variables such as physician training and experience, clinical setting (community vs. tertiary care center), and overall comfort with use of the OAR. We were unable to conduct a subgroup analysis on children under 6 years of age due to lack of data. Some studies included all patients <18 years, while others included only those between 6 and 16 years13 or between 5 and 19 years.12 Only two studies4,23 provided data on the number of patients in the younger age group; 3 of 35 patients (8.5%) in one24 and 81 of 761 patients (10.6%) in the other.4 This poses a significant problem; clinical experience suggests that the younger patients are more difficult to assess. Younger children are not always willing to ambulate on command, especially if they are in pain, nor are they always able to coherently verbalize how they feel. As a result, we cannot be confident of the accuracy of the OAR in children who are less than 6 years old and recommend caution when applying the OAR to this age group.
The types of fractures considered significant and therefore included in the individual study sensitivity and specificity calculations varied. Some studies included all fractures, some only fractures ≥3 mm, and others omitted SH-I and fractures that were termed “insignificant fractures.” Whenever possible, this analysis included all reported fractures in the calculations.
Finally, in eight of the studies the OAR were applied retrospectively to data collected at the time of assessment and later compared to the criterion standard x-ray data. This may have introduced bias based on prior knowledge of the diagnosis and interpretation of chart information. Authors attempted to minimize the potential for bias in seven of the studies by blinding the investigators to the radiologists’ diagnoses. However, we do not know if this was successful.
Based on the findings from this study, the OAR appear to be a reliable tool to exclude fractures in children greater than 5 years of age, presenting with ankle and midfoot injuries without worry of missing a significant fracture. Based on the pooled data, application of the OAR in children would result in a 24.8% reduction in x-ray usage.
- 5Reappraisal of use of x-rays in childhood ankle and midfoot injuries. Emerg Radiol. 2002; 9:88–92., .
- 24Validation of the Ottawa Ankle Rules for varus ankle trauma in children: a prospective study of 160 cases [abstract]. J Bone Joint Surg. 2004; 86-B(Suppl 1):20., .
- 25Application of the Ottawa Ankle Rules in children: a validation study [abstract]. Acad Emerg Med. 2000; 7:584., , , .
- 28Applicability of the Ottawa Ankle Rules in an urban pediatric population [abstract]. Acad Emerg Med. 1999; 6:452., , , et al.
- 32The use of the Ottawa Ankle Rules in children: A survey of physicians’ practice patterns. Can J Emerg Med. 2008; 10:255., .
Data Supplement S1. Terms for search strategy.
Data Supplement S2. Inclusion/exclusion criteria.
Data Supplement S3. Data extraction form.
Data Supplement S4. Methodological quality.
Please note: Wiley Periodicals Inc. is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
|ACEM_333_sm_DataSupplementS1.pdf||9K||Supporting info item|
|ACEM_333_sm_DataSupplementS2.pdf||74K||Supporting info item|
|ACEM_333_sm_DataSupplementS3.pdf||155K||Supporting info item|
|ACEM_333_sm_DataSupplementS4.pdf||40K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.